PROOF - You CAN rank down & lose SR with Positive Win Ratio

Yeah but that’s the thing it’s not important how well I do relative to the average player stats. In the same way that medals are not a good determining factor of value neither can I trust the pbsr to accurately know how much value I’m contributing to. Get to remember it’s not a human and this is an abstract complex idea of value, something which you can never trust an AI too comprehend especially not within the budget of this video game they won’t create an Al upgrade enough complexity to know that just to judge gameplay

What is show statistically according to the AI pbsr is that I have not improved one single bits so I played which is untrue I’ve improved a lot. Because I have close to a thousand games my windracial would be indicative of my skill which would show that I’m at least average I’m contribution to my games and with those games I have more wins than losses so I should have ranked up. Pbsr cannot judge me to be anything less but it does anyway. Or maybe it’s the amount of Smurfs in the lower ranks that is saturating the stats to be higher than they should be at my rank who knows regardless I don’t want to be held back because the system assumes I’m playing bad based off some calculation does when I’m not actually playing bad, that’s what wins and losses determine. An extra system just makes things more prone to error.

Getting lower in SR and improving over multiple seasons are not mutually exclusive. Everyone that keeps playing is getting better all the time. The average player (SR 2350) is a constantly moving target, always increasing in skill to stay in the same place. SR is relative to all the other people still playing this game. PBSR isn’t an absolute measure of your statistics, it is relative to all the other people at a similar rank to you. If they keep putting up higher numbers, you have to outpace their increases to merit the adjustments upward.

Having an overall positive win rate but being lower in SR just means that you help teams win in ways other than the statistical measured ways. If you weren’t having that impact, then your SR would likely be falling faster, because the win rate would be negative.

Instead of finding fault with the algorithms, if you are interested in climbing, you should determine ways to play the way that the algorithm encourages while still winning.

2 Likes

Yeah that’s what I figured although lately I’ve been measuring my SR and I seem to be getting above average on most of them and losing less than normal. Perhaps this wasn’t always the case but it seems to be the case now at least. I’m not too fond of the idea of having to play according to an AI just to level up better though I wish pbsr wasn’t a thing in the first place but I guess it is what it is

What OP is showing proof of, is that he is playing terribly, but his team is carrying him

1 Like

And where’s your proof of that, do you know how pbsr’s calculated exactly? I’m providing the variables that I can account for, you trust an AI to be able to do a human’s job in measuring value. No way I has beat the touring test the way I can imitate human, you really think that blizzard develops some super advanced AI for a game to calculate value?

A win ratio after close to a thousand games should be pretty reflective of my own personal Performance. In the end I got 52 win percentage due to me failing at everyone but brig. However a 52% win ratio is still slightly above average so if pbsr was calculating my contribution correctly which by now should reflect on my win ratio, then pbsr should not have rated me either above average or below average. But that’s not the case do you think of it close to a thousand games won’t reflect my own skill? Because what I’ve seen said on this forum is 100 games and you could get an idea of your skill we’re talking close to 1, 000.

Also blizzard did not want to release the methods which pbsr is calculated because they feared it would be abused, however if pbsr was did its job correctly, then it would not be able to be abused because it would be measuring your skill which cannot be abused unlike stat farming. Placing trust in the pbsr system is therefore equivalent to placing trust in the medal system to calculate your value.

There’s enough information to have a valid argument for my side but yours is just an assumption.

As people have previously mentioned, your stats are lower than the average of the heroes you play

2 Likes

Here’s the thing, win rate has nothing to do with pbsr, some boosters were talking about how to maximize SR gains and have come to the conclusion: On fire% is the most important, kill participation, E/D ratio, elims/10, deaths/10, and accuracy/damage/healing are actually on the bottom.

So if this is the case, pbsr is being calculated just fine for you, and once again, win rate doesn’t affect it.

2 Likes

Yes and how does having low stats guarantee low value or playing terribly? Especially on the character like break who is primarily specializing around CC? Also are you sure that the stats it does measure are important to the game that I’m currently playing? Does it know that this game perhaps valued more offensive DPS than healing? The logic your pushing for filled the same fallacy that believing that believing the medal sys accurately measures value

Alternatively one could use Mercy and just healbot tanks exclusively and wind up with above average stats. I it doesn’t necessarily mean you did better than the person who didn’t heal bot .

PBSR is calculated however it’s calculated of that I have no doubt, however is the intent of Blizzard was to make it appear like it would help you rank up if and would soften the blows of loss to help you rank up and it’s assumed that it would do the opposite in the opposite case. To the original intention was that it’s suppose give an accurate assessment of your player skill and uses that to further influence your.

PBSR inherit has nothing to do with win ratio however after enough games one could use their win ratio to determine their overall contribution they brings to the to the game which can be determined by the win ratio. In this case a percentage of 52 puts me slightly above the average player in terms of wins which in turn means at the very least my player skill skill places above average, and as you can see there’s only one season where I actually get a 45% ratio.

So if the sheer amount of games played combined with my win ratio are showing that my player skill slightly above average. That contradict PBSR which this far has been calculating my player skill has been low average thus the SR losses. This contradicts. And whereas win ratio only gets more accurate determining skill with the higher number games you play (884 is rather high, it’s often recommended on the forms you simply play 100 to get a good idea of your skill), PBSR calculations can either be guaranteed for every character and are prone to errors because AI detection is not sophisticated enough to accurately measures, which is why they never wanted to reveal exactly how those calculated, due to the possibility that it would be abused (which could not happen if PBSR was an accurate measure yourbskill because you cannot farm skill but you can farm stats)

Once again, your win rate has nothing to do with pbsr, just because you have a positive rate, doesn’t mean you should be climbing (500-2999), you can consistently under perform and still have a positive win %, but if you lose more SR per loss than you gain per win, you won’t get ahead, like wise if you gain more SR per win you can climb with a negative win%.

The fact you haven’t climbed (I see you have gained a bit now) shows your stats were too low, thus losing more SR than you win.

3 Likes

What makes you think it’s a super advanced AI? Any number of machine learning methods would suffice–if it’s even required at all.

In any case, you’re being “graded” on the same “test” that everyone else is.

Win ratio doesn’t reflect personal performance under any elo system. Win ratio relative to who you’re winning against reflects personal performance.

Your question rephrased: How does scoring below average on an exam 10x mean you don’t know the subject?

You don’t get “on fire” by heal botting with Mercy. So no, your stats would be mediocre or below average.

2 Likes

You can consistently underperform and have a positive win ratio in the same way you could roll a six on a die six times in a row despite there being 1/6 of a chance. This however becomes less and less likely the more games played. Anyone familiar with averages over time knows that no matter how volatile the beginning the results, if done over a sufficient amount of time they will pan out. And so since there exists the chance of whether I get the good teammates or the bad teammates it’s a 50/50 so their effects on my win ratio over the long run become redundant. After 884 games, over time the likelihood that I’m underperforming is slim to none, according to the effects of averages over time. Now you keep saying that I’m underperforming. And that is purely based on faith on what grounds exactly, enough that you would call this mathematical logic redundant?

Also why do you say that I shouldn’t be climbing if I have a positive win ratio, in any other competition that dictates a climb, however you’re placing your faith in pbsr and I’m wondering why as I’ve given to you the reasons why doing this would be as reliable as trusting medals to be a good judge of your skill. Also consider what if your team performs really well? So well that they are getting the picks before you can, not because you can’t get the picks but because they are better than you. In this case you are not necessarily bad but because your teammates are a lot better, to an AI system you will look bad by all possible statistics fed to it. It will assume you are bad because you got outplayed by you own team, an odd punishment. The inherent problem with AI it is not capable of human understanding, I’m not sure why you place your faith such a system, whereas something like win ratio is mathematically proven and continually becomes more accurate the more results you feed into it

The biggest hurdle in AI is abstract understanding because it is coded purely logically, as a human you probably take for granted or do not even consider certain things because we do it so easily.

Consider a Symmetra teleporter behind the enemy team, how does the game know whether the decision to place the teleporter behind the enemy was impactful or not ? As far as all the statistics say all you did was set up a teleporter in your team went through it. Were the resulting deaths that possibly happen afterwards the result of the surprise of the teleporter trick or of the player’s skill if so how much was due to the player skill compared to the decision to use the teleport and how do you calculate that with the information fed to the machine learning program? What if the people teleported possibly died how much was it you fault for not placing the turret more strategically? The amount of information the AI would need to be fed and the amount of calculations and considerations it would have to take into account would be ridiculously complex for something which maybe to us is extremely simple.

Of course this is rephrase to work to your advantage, on pen and paper it would appear so, but everybody knows that saying about it looking good “on pen and paper”, because if this was your driving exam would passing it with a flawless score guarantee that you’re a good driver? It would not because in practice there are more things to consider then this basic information. Even if you knew that you had to stay in your lane, if a car was driving the wrong side of the lane you would have to break that rule in order to avoid it. Same reason why your team having the highest stats doesn’t guarantee you’ll win.

Assuming your rank was always perfect at all times. All you need to be familiar with is average is over time to realize the redundancy of this argument. Yes you might be facing players who are lower skilled and lose more Sr on the loss but you are also playing players who are higher skilled and lose less SR on the loss. Same thing goes with the winsz if you are winning against lower ranked players you will win less Sr if you’re winning against higher ranked players you will win more Sr these essentially cancel each other out no matter which way you do the math because averages over time make it redundant. Besides pbsr isn’t calculated by the enemy teams.

Natural selection is based on win ratio. Time and time again the species with the highest win ratio/survival ratio prevail for a reason. This is the way that winners are calculated by the universe itself. What ELO systems do is try to accelerate processes which usually takes a while, and attempt to speed it up based on information we have. We may be able to make some accurate predictions using our current mathematic and knowledge and this may be done to things like weather which is a lot more predictable but even then complex to predict and not perfect, but it is not so easily done with people. if it could be done, all sports would be predictable because you can just take their pbsr and calculate their value and the team with the highest pbsr would guarantee win.

So then how do you get on fire with Mercy? And if you don’t get on fire by healbotting does that mean that healing has no value? They could legitimately be times where heal botting is the most valuable thing to do.

1 Like

Do not try to math this, NOBODY has the accurate numbers.

And yes, you can have a positive win rate and still under perform, the same way I have a negative win rate and still climb (albeit slowly) because I have the stats. After 844 games, between you slowly improving, and the other players doing the same, your stats have not moved, we’re all better than when we started, the problem is, others improve faster and leave us behind. It’s the hard stuck players that DO NOT understand this point.

Medals are worthless, lets start with that. And yes I place my faith in PBSR, since I’m not losing rank thanks to it. Let me ask you this, before we continue this conversation any further: Do you find the games you play, on any given role, easy, close, stomped?

2 Likes

The algorithms are designed by humans and we know how they work.

You get on-fire points if your teammates use it. Did you send 3 turrets through that do damage and kill a support? Did everyone on your team use it?

The thing is, every valuable play or action that you’re performing is also being performed by other players. The difference is, they also have better stats.

Can you explain the value that you’re contributing that higher elo players with better stats are not contributing?

It’s like scoring below average on an exam and then saying the exam was unfair because you should get points for showing up, spelling your name right, answering all the questions, using the correct pencil, and finishing in the alloted time. Everyone in your class is doing the same thing but some people are answering more questions correctly and therefore getting more points.

Student A scores in the 99% percentile of the class over 10 exams.
Student B scores in the 45% percentile of the class over 10 exams.

If you had to guess, which student knows more about the subject?

Win rate != SR. It doesn’t matter if it’s over 10 games or 10,000 games. You must necessarily consider who you are winning against. There’s nothing redundant about that.

A 2800 FIDE chess Grandmaster clapping a 1600 rated player for 100 games is going to have a 100% win rate. Their rating will barely move for winning, but it will drastically drop if they lose.

I study human genetics and evolution for a living. You’re analogy is wrong and you don’t want to go here, I promise.

Healing, damage boosting, resurrections, not dying, objectives, assists, and elims. Healing alone will not fill the on-fire meter for Mercy.

Damage boosted killing blows, resses, and objectives give the most on-fire points.

Mercy gets 1 on-fire point per 5 HP healed. Outside of Valk, she heals for 55 HPS, meaning every second she can get 11 on-fire points. Points degrade at 5 per second for a net of 6 on-fire points per second from healing. You need 250 to be “on fire.”

That means you have to “heal bot” for 41.66 seconds without interruption to be on fire i.e.–it isn’t realistically happening.

2 Likes

The thing is you can math it, as I’ve done with the wind ratio that’s legitimate math right there that’s legitimate reason and you have not disproven did a higher win ratio after 884 games would be now be a reflection, and that is needed to at least that your assumption is of very low probability.

If you did climb with the negative win ratio that doesn’t actually help your arguments if anything didn’t reinforces the fact that there can be a problem with PBSR. While win ratio will always remain calculable, is simple, and it’s significance is undeniable for anyone who knows time. The same cannot be said for pbsr and mysterious system, and is a man-made concept and system that is far more complex and more error-prone while having over 50a% in a ratio will always signify more or an advantage for anything it is simple and universally true which is applicable to the very laws of nature and to any universal truth that would show one is slightly better than the other.

Still you have not addressed the other points I’ve made, picking and choosing the points you argue while ignoring other is not how you prove yourself to be correct, that’s only manipulative way to attempt “win” an argument.

Edit accidentally submitted it I’m currently editing it to include your other points

So, do you find your games easy, closely skilled, or stomps? Answer this.

2 Likes

When you stomp you earn very few points. When you get stomped, you lose so many points.
Is this supposed to be fair?

That’s a mixed bag, and wouldn’t even matter some games I found easy and we should have won and we didn’t, put some of those games we did. Some games I found hard and I thought we would lose and we didn’t, but then some games we did.

If you going to die further into this conversation line that is going to put us in the field of heavy assumptions and if we’re going to argue based on our assumptions alone we’re going to have a very long argument

I’m very curious to see how you think this is going to help.

Alright, so it’s mixed, at your current rank all matches are between easy and stomps, next question: Do you win more SR than you lose?

2 Likes

That’s not even what I was talking to about at all, where did I say that we don’t understand our own algorithms? I said AI has difficulty with abstract understanding because our algorithms are logical and not abstract in nature. It is purely the difference between meaning and logic. Meaning is not logical and therefore it’s difficult find a way to describe abstract concepts. An example how would you program a robot to know to truly know the meaningfullness of Vincent Van Gogh’s Starry Night paintings. Sure you could “hard code” the reasons such as “we like it because it’s surreal” however let’s just the AI reciting information not understanding on it’s own. Do this method you would have to predefine almost every possible situation in the game along with its rated value which would be an immense undertaking…

Why would you get on fire if your teammates use it it’s not guaranteed they’re going to get value from it or not?

If you sent three turrets to get a kill then of course this is no different from shooting a bullet. However it would only know its value in terms of you getting a kill and it would award you that. It cannot however easily know if that support was particularly hard to kill because you couldn’t get into the back line. Therefore it does not differentiate it between getting any other type of kill, and therefore it would assign the value of the kill to be the same as any other kill. Even accidentally killing a player would be counted as your actual skill because that’s.

And what would everybody on your team have to do exactly with how much value it was everybody could use it and die everybody could use it and succeed everybody could use it and that made no. It doesn’t know how well you used it only that you used it

Valid point we are on both sides contributing equal. However there’s value that could be interpreted in simple information and there’s value that requires context. This is why getting gold medals means that there’s a pretty good chance you were doing okay but it is not a guarantee, and why you could still win matches even though your team got the lowest stats that unknown that is not shown in metals or statistics is what you could possibly be doing more than the obvious medals and statistics that are being. Something like pooping away an enemy hog after he hooked your teammate will only be interpreted by the system as you pooping a hog and will value it for the damage it caused. There is no information however saying that you just saved your teammates life.

Another example on the map hanamura when you are on the 2nd point in the wooden room. There is a bridge in the entrance that you must cross to get to the platform. Sometimes during the enemy push I like to go off to the side and boop one of the tanks off the bridge because it messes with their push, and they essentially are pushing with one less player now than they had planned while their teammate has to make his way up back the stairs to get to the fight during this time we could capitalize on it and we could have a numbers advantage. This is can be very valuable , this type of value will not be recognized. So in this case my playstyle centers around CC which I do like to capitalize on with brig for value, another brig player my capitalize more on keeping her inspire up or damage even, and another could prioritize keeping the other healer alive and gain value that way. So with many different venues to get value, and some even being incalculable, the possibility that it is not calculating it correctly is very real.

I never denied that one would likely know more about the subject. Obviously the one with 100% would likely know more However knowing more and placing it into practice is two different things. Knowledge is power only when knowledge is applied correctly, without proper application it is just useless information it is just information.

Similarly, I could watch one of your replays and I bet I can point out mistakes and point out what you should have done, and just because I pointed these things out doesn’t necessarily mean that I’m better than you

And in the case of this being a driving exam which you’ve ignored, again you could score 100% on your exam it does not mean you’re a good driver, it just means you memorized information well.

You must have not read what I’ve written. I clearly pointed out why it’s redundant, unless you can take what I’ve said and counter the logic behind it showing me why it’s not valid then you’re just reiterating your old point.

I’d welcome the challenge because basically you could reword it many ways but win ratio does apply.

Species A versus species B they both have 50 children due to high mortality rates. These 50 children are in direct competition with each with a genetic variant from the other batch. One of these batches does not survive the other has two remaining children that do survive. Let’s say species b survives that is one win for species being and zero for a. However that’s just one family if we consider 100 families on both sides their survival ratio will be pretty much like their win ratio. And a higher win ratio would mean that species b is likely more fit to survive as shown by the results.

If healing alone will not feel the on fire meter, what happens when the most valuable thing to do is healing alone? Wouldn’t this essentially rob you of the value you should have?

Remember, this does not apply to Diamond and above. You get the same amount of SR for beating a team 200SR below you as you do beating a team 200SR above you, so in that rank, relative skill doesn’t directly affect your SR.